FullFact / health-misinfo-shared

Raphael health misinformation project, shared by Full Fact and Google
MIT License
0 stars 0 forks source link

73-write-an-evaluation-script-using-promptfoo #83

Closed dearden closed 2 months ago

dearden commented 2 months ago

Fixes #73 .

Adds some basic files which allow us to compare prompts with promptfoo in a realistic environment.

The example used was comparing prompts with and without context.

The tests themselves (i.e. the means of comparing) have not been implemented as part of this PR. They are to come in future updates (#80 #81 ).

But the generic scripts here mean we now just have to write the individual functionalities for each test and they can slot in.


Pull request checklist